Sensitivity Analysis of Model Output: SAMO 2004

The SAMO conferences are single-track meetings that promote in-depth technical discussions of issues related to the uncertainty and sensitivity analysis of model predictions. In the context of the SAMO conferences, uncertainty analysis refers to the determination of the uncertainty in model predictions that results from uncertainty in model inputs, and sensitivity analysis refers to the determination of the contributions of the uncertainty in individual model inputs to the uncertainty in model predictions. Here, model is used broadly enough to include large analyses that involve multiple individual models that are linked together, and model input is used broadly enough to include designators for alternative models or modeling assumptions. Topics considered at the SAMO conferences include innovative methods for uncertainty and sensitivity analysis, experimental design and sampling plans for computer experiments, model calibration and validation, probabil-istic and nonprobabilistic structures for the representation of uncertainty in model inputs and outputs, and decision-making under uncertainty. The uncertainty and sensitivity analysis of model predictions is important in all areas of modeling and specific modeling topics that have been considered in the SAMO conferences include problems from economics, engineering, ecology, nuclear safety, and physics. The novelty of simply being able to carry out a complex calculation is past. Now, the relevant question with respect to an analysis is not ''What is the answer?'' but rather ''What is the uncertainty in the outcomes of the analysis and what gives rise to this uncertainty?''. Or, put another way, ''How much confidence should be placed in the results of the analysis?''. Answering these questions requires the performance of an uncertainty and sensitivity analysis and is essential to drawing properly informed insights from an analysis. Large numerical analyses are now often used to support important societal decisions. Appropriately informed decisions cannot be made on the basis of such analyses without knowledge of the uncertainty associated with their results [1–16]. The SAMO conferences provide a forum for the presentation and discussion of both techniques for and results obtained in uncertainty and sensitivity analyses. By providing this forum, the goal of the SAMO conferences is to encourage and enhance the use of uncertainty and sensitivity analysis in the assessment of model predictions. In support of this goal, the SAMO conferences have a tradition of publishing an associated special journal issue [17–20]. Consistent with this tradition, the present issue of Reliability Engineering and System Safety contains selected papers from the SAMO 2004 conference. …